Please note neither GPTZero, nor the ChatGPT detector are 100% accurate and it is expected that as every iteration of any AI bot is launched (eg ChatGPT4 is due later this year), AI detectors will also likely need to be upgraded. So, this appears to be an “arms race” that may never end, given that future AI writing tools will be trained to produce less detectable content. Moreover, if we take the view that detection tools may even be overtaken by new text generators that mimic human variation in language, we will need to rethink assessment. (Here are two examples: Thoughts On Academic Integrity and AI Technology, Two Simple Things To Change In Your Assignments To Make It Harder For AI Writers: Specific Prompts And Personal Experience Prompts)
Here’s an interesting article:
Snippet:
Are AI detectors reliable?
No, at best they are indicative. Published claims to reliability vary greatly, between about 26 and 80%. In other words, expect them to be wrong between a fifth and three-quarters of the time. It is possible they will improve, but this should be viewed as an arms race rather than a stable situation. For instance, recent advances with ChatGPT show that it is possible to coach it to write with more complexity, making it harder to detect.
If you are going to use these tools, we recommend that you check your sample with more than one.
